jacobpolloreno/tensorflow_serving_hug_api

A hug api that handles requests and responses from a tensorflow serving server(companion repo)

Machine Learning Demos

LIVE: demo.jacobpolloreno.com

Current Deployed Models

• Cifar 10

Description

This is the REST API that sits between the frontend server and Google's Cloud ML, where my personal ML models are hosted.

It uses a Python API called Hug which facilitates the Google API client request. After the user inputs data into the requested model, the server will then do some of the preprocessing for the model such as image resizing and converting images into b64. A request is then sent to the model and a response received with labels and probs. The server then sends that data to the front-end server to display to the user.

• Code for the front-end can be found here. It uses a python web server and Google's api client.
• Code for the custom model(s) can be found here. They are built with Tensorflow and saved to be used on a Tensorflow Serving host(i.e. Google Cloud ML Engine)